►
From YouTube: 9-18-2020 - Node.js N-API Team meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Welcome
to
the
node.js
napi
team
meeting
for
the
18th
of
september
2020.
we'll
follow
our
standard
practice
of
looking
at
what
we've
got
tagged
for
milestone
11.
before
we
do
that.
Does
anybody
have
any
announcements
they'd
like
to
share.
A
I
guess
one
that
would
be
good
mentioning
is,
you
know,
nikola
is
working
on
a
release
now,
so
we
expect
a
new
release
of
node
add-on
api
and
I
guess
the
next
day
or
so
right.
B
Yeah
yeah
yeah
yeah.
We
are
near
to
to
release
the
the
version
3.0.2
for
loaded
on
api
and
yeah.
Finally,
we
can
solve
the
problem
that
we
introduced
for
on
windows
yeah
we
bracket
this
week.
We
create
some
breakage
on
on
on
windows
and
yeah.
B
Now
we
finally
solved
this
and
we
added
the
new
api,
the
add-on
class
that
yeah
help
to
handle
the
loading
of
the
native
dawn,
specifically
multiple
lords,
in
the
same
trend
and
the
lord,
the
lord
of
native
dawn
in
multiple
threads,
and
we
added
some
tests
and
fix
on
on
on
all
the
the
api
and
documentation.
B
So
this
is
what
what
happened
with
the
next
release.
A
Sounds
great
thanks
for
leading
that
effort
and
as
well
as
creating
the
tests
for
the
the
new
piece
of
functionality,
so
that's
definitely
important.
Okay,
let's
move
on
to
the
issues
tag.
The
first
one
is
discussion
on
api
7..
I
think
we've
covered
that,
but
I've
seen
some
recent
stuff,
so
I
think
gabriel.
Do
you
want
to
bring
that
one.
C
Yeah,
yes,
I
wanted
to
bring
up.
Let's
see
what
we
have
here.
I
some
stuff
has
happened
since
I
went
to
bed
okay,
so
we
could
add
okay,
okay,
all
right,
I
see
so
so
you
said:
you're,
okay,
either
way
yeah.
So
we
had
a.
We
had
a
pr
come
in
that
that
basically
asks
to
be
added
to
an
api
7.
But
since
it's
a
new
api,
well,
it's
not
really
a
new
api.
It's
just
a
couple
of
extra
values
for
hey:
where
did
where
did
the
pr
go?
C
Oh,
yes,
it
landed,
yeah
yeah,
so
so
adding
a
couple
more
property
values.
So
basically
we
could
release
an
api
7
without
it
or
we
could
wait
until
until
we
managed
to
backport
them,
because
it's
a
really
really
tiny
change,
so
it
should
be
easy
to
back
port
and
and
then
release
an
api
7
with
the
properties
intact.
So
it's
not
really
new
api.
C
C
That's
true
yeah
yeah,
yeah,
you're
right
it
is.
It
is
some
reminder
right,
but
basically
so
shall
we
shall
we
proceed
with
an
api
7,
as
is,
is
the
question,
or
shall
we
wait
for
for
for
this,
this
small
pr
to
trickle
back
and
then
and
then
tag
an
api,
seven.
A
A
It
doesn't
sound
like
it's
urgent
and
you
know.
Maybe
it
goes
out
in
napi,
eight,
as
part
of
you
know,
16
or
something.
C
C
That
was
it
so
okay,
so
then
I
will
proceed
with
landing
that
pr,
the
the
one
that
that
establishes
seven
and
then
start
trickling,
that
back.
A
And
I'll
guess
I'll
add
it
also
sounded
like
the
the
original
author
was
happy
to
wait
too
on
that
one.
So.
A
Okay,
next
one
is
review
external
examples.
I
don't
think
we've
made
any
progress
on
that
one.
No,
unfortunately,
the
next
one
is
the
tracking
issue
for
modules.
We've
noticed.
B
Okay,
so
I
I
know
I
just
saw
an
issue.
B
On
okay
on
sqlite.
A
A
B
B
B
B
I
I
I
won't.
I
won't
find
the
link.
A
B
D
E
B
Okay,
so
I'm
posting
on
on
the
chat
the
because
the
the
issue
was
opened
on
on
on
on
not
keep,
and
essentially
yes,
it's
it's
true
that
we
did
didn't
back
ported
this
on
on
note,
10
and
on
on
sqlite.
B
Not
be
built
version
yeah
to
on
their
binding
jeep,
so
that
I
will
obtain
this.
This
error
on
node
10,
because
yeah
in
no
time.
A
Right
yeah,
I
guess
the
question
is:
was
there
a
reason
we
didn't
backport
it
or
we
just
didn't.
B
B
The
outdoor,
the
outer
of
the
of
the
the
the
issue
closed
because
he's
witched
to
a
new
version
of
load
and
now
it
compiles
and
and
it's
okay,
but.
B
This
kind
of
problem-
maybe
we
need
to-
I
don't
know
if
we
need
to
do
something
on
documentation
or
I
I
know
idea.
I
just
saw
this
this
issue
right
and.
A
A
A
B
A
A
B
B
A
Okay,
so
I'm
gonna
go
back
there.
That's
also
where
we
usually
just
take
a
quick
look
at
it.
Doesn't.
C
A
I'm
wondering
if
we
should,
it
might
be
time
to
write
like
a
blog
post,
that
we
could
all
collaborate
on
in
terms
of
the
experience
over
the
last.
You
know
over
the
last
few
years.
Would
that
make.
B
B
Yeah
and
what
do
you
want
to
to
talk
yeah
to
on
these
and
these
these
blog
posts
so.
A
A
A
B
Okay
and
how
okay,
how
do
you
want
to
write
these,
these
blog
posts
so
using
a
google
doc
or
maybe
we
can
put
a
some
ideas
on
the
the
issue
and
that
then
we
make
together
all
the
things.
B
Yeah
for
me,
it's
it's
a
good
option.
A
A
I
haven't
had
any
chance
to
look
at
enabling
debug
testing
for
add-ons
nothing
new
on
that
one.
Let's
take
a
look
at
the
list
of
issues
so
actually
here
we
just
want
to
go
to
no.
B
Yeah-
and
I
think
that
these
is
done
because
yeah.
A
A
B
B
Yeah,
I
remember
that
cabin
was
working
on
these,
but
I'm
not
really
sure.
A
A
Some
right
so,
okay,
so
I
think
not
anything
too
too
new.
Let's
look
at
pull
requests
if
there's
anything.
B
Yeah
there
is,
there
is
a
torrent,
yeah.
B
Yeah,
I
think
it's
a
leak
of
transit
data
right
yeah,
you
lost
calm
and
I
was
about
to
take
a
look
yeah.
F
E
C
Data
yeah,
I
think
I
think
I
may
have
actually
noticed
this
as
well-
that
we
weren't
deleting
that.
So
I
think
this
is.
This
is
pretty
much
good
to
go.
C
Yeah,
it's
kind
of
it's
kind
of
hard
to
it's
kind
of
hard
to
to
to
test
this,
though
you
know
like
these
leaks,
because
I'm
thinking
it
would
be
nice
if
we
had
a
test
for
this,
but
but
there's
really
no
way
to
do
it,
because
it's
an
internal
structure
and
we
have
we
have
no,
we
have
nothing
informs
us
that
it
was
deleted.
You
know
so
that
we
can
fail
if
it
wasn't.
C
F
A
C
That
checks
for
a
mem
leak.
A
C
Oh
boy,
okay:
well,
we
we
do
have
like
a
uv
handle
leak.
We've
got
that,
but
I
don't
know
about
a
memory
leak.
C
C
We
even
have
one
in
add-ons
which
is
nice,
so
so
in
the
non-napi
add-ons
we
have.
We
have
something
called
uv
handle
leak
that
might
be
interesting
right.
A
C
A
A
Perfect,
okay
and
go
ahead.
A
B
Yeah,
I
just
saw
that
the
the
cia
busted
now
and
yeah
after
the
meeting
I
will
publish
okay,
give
me
some
some
meals.
A
A
A
A
C
C
D
C
Well,
I
mean,
as
it
is,
okay
as
it
is,
it
should.
C
C
So
because,
right
now,
since
it
since
it
is
zero
right,
then
then
those
those
who
might
wish
to
use
it
have
written
their
code
in
such
a
way
that
they
don't
use
it
because
they
are
unable
to
use
it
so
they're
like
oh.
I
can't
use
that
here.
Okay,
so
then
I'll
just
do
something
else
right,
so
so,
presumably
right!
So
then,
if
if
it
is,
if
it
becomes
all
of
a
sudden
usable,
then
none
of
the
code
that
would
have
used
it
but
couldn't
should
break.
A
C
They
use
it,
so
they
didn't
right
right,
so,
presumably
fixing
this
would
not
break
anybody,
but
would
allow
people
to
rely
on
it
as
they
did
with
pure
js,
without
even
knowing
that
this
was
ever
a.
C
C
C
C
Press
null
yeah,
but
if
you
pass
in
null
then
then
then
you
we
should
retrieve.
So
basically
what
we
should
do
inside
the
native
function
is
is,
is
what's
what's
happening,
what's
happening
in
javascript.
So
if
you
look
at
what
steps
will
reproduce
the
bug.
C
It's
like
so
you
retrieve,
you
basically
call
execution
async
id
from
from
from
the
asynch
hooks
model,
and
you
expect
that
back
so,
basically,
we
should
also
be
calling
execution
async
id
from
the
async
hooks
model,
add-on
or
module,
and
so
actually
makes
sense.
C
A
F
C
A
A
F
A
A
A
A
B
A
A
A
That's
good,
okay!
Well,
I
think
that's
it.
Let's
go
back
to
here.
We've
looked
at
the
burn
down
the
issues,
and
I
yes
there's
this
one
too.
So
this
is
potentially
it
sounds
like
in
canary
one
of
the
tests
is
failing,
so
we're
getting
this
test
object.
Crashing
last
thing,
I
tried
to
clear
the
workspace:
let's
see,
if
that
actually
had
any
different
made
any
difference,
because
we,
the
problem,
is
it
didn't
recreate
locally,
which
was
annoying.
A
Basically,
both
myself
and
targos,
compiled
it
on
different
machines
and
it
didn't.
I
think
we
want
test
v8,
v8.
A
A
A
A
C
I
I
checked
this
one,
it
it
it.
At
least
the
latest
santos
ran
out
of
memory,
so
yeah.
If
you
go
to
not
santos,
if
you
go
to
fedora
latest
and
I
checked
check
the
console
for
that,
it's
not
a
test
failure.
It
just
ran
out
of
memory.
Oh
wait!
What
I
looked
at
something
else,
then
I
just
closed
the
tab.
Like
that
see
what
I
was
looking
at.
A
Seems
like
yeah,
so
basically,
we've
got.
You
know
we
somehow
have
to
be
able
to
recreate
it.
C
How
do
I,
how
do
I
try
to
recreate
this
locally.
C
A
C
A
A
C
A
F
C
A
Even
running
like
all
of
the
the
add-on
like
the
the
napi
add-on
test
together
in
a
loop,
you
know
and
like
100
of
those
didn't
make
any
difference.
So
it's
curious
as
to
why
we
can't
recreate
it
and
it's
100
seems
to
be
100
in
the
build
so
anyway
yeah.
If
you
can
try
the
fedora
one
and
otherwise
we'll
have
to
sort
of
brainstorm.
What
else
we
can
do
so
yeah,
I'm
gonna
put
that
at
the
top,
so
we
don't
forget.
A
A
Okay,
we
were
looking
at
and
I
think
that's
all
right.
I
think
we're
saying
that's
about
it.
So
anything
else
people
want
to
discuss
before
we
close
up
for
today.
C
C
And
basically
we
are
at
the
point
where
we
wish
to
create.
We
wish
to
compile
web
idl
such
that
it
outputs
like
an
api
calls,
rather
than
v8
or
or
some
some.
A
C
B
I
remember
that
there
was
a
guy
roman
dev.
I
remember
the
nick
on
yeah
on
other
dawn
api,
that
if
I
were
a
member,
did
something
with
web
by
dl
and
and
nodded
on
api
yeah
to
create
the
idea
drone
and
you
can
you
can
search
on
not
that
done
api
repo
for
sure
you
will
find
an
issue
where
he
describes
what
he
did.
C
A
Jim
jim
did
do
some
work
on
something
that
would
generate
an
api
not
from
web
idl.
I
think,
but
from
other
things
so.
A
Can
you
just
give
me
sorry,
go
ahead.
B
Yeah
another
project
that
yeah
and
they
they
opened
an
issue
on
another
done
api.
B
I
don't
remember
the
name
so
yeah
I
I
I
can
send
I
I
can
search
later
and
I
send
you.
I
will
send
you
by
me
gabriel.
C
C
yeah
it's
just
kind
of
stalled,
but
yeah,
so
he
worked
on
it,
but
then
he
didn't
post,
didn't
post
any
any
results
and
and
and-
and
somebody
asked
last
april,
if
there's
been
any
progress
made,
then
that
was
it
yeah.
There
are
something.
B
But
he
made
a
presentation
about
about
about
it,
so
I
think
that
even
if
it's
a
prototype,
so
you
you
can
see
to
to
yeah,
extract
some
ideas
right.
C
A
C
Yeah,
basically,
yes,
it
would
take
care
of
like,
like
a
type
validation,
that
kind
of
thing
and
then
it
it
would
not
be
a
complete
solution
right,
because
your
your
native
api
does
not
necessarily
have
exactly
the
same
interface
as
the
javascript
one
and-
and
you
still
have
things
like
you
know,
pointer
ownership
and
and
and
how
do
you
do
callbacks
and
the
the
the
usual
the
usual
suspects
that
that
make
complete
generation.
A
C
Is
right,
yeah
yeah,
so
so
so
basically
I
mean.
Even
if
you
had
a
one-to-one
mapping
you,
you
would
still
run
into
things
like
pointer
ownership
and
that
kind
of
thing
right.
So
the
idea
is
not
to
completely
solve
the
the
issue
of
of
of
exposing
a
library
to
to
javascript,
but
rather
to
solve
the
issue
of
of
having
to
type
a
pile
of
code
just
to
do
basic
stuff
repeatedly
right
and
then
leave
you
to
to
do
the
actual
thinking
and
and
heavy
lifting
of
you
know,
pointer
management
and
and
reference
management.
A
C
Well,
in
the
case
of
web
gpu
and
api,
I
spoke
to
to
the
person
driving
the
spec
during
the
meeting
and
he
said
that
it
was
a
pretty
chatty
api.
So
I'm
thinking
core
because
even
core
has
a
significant
overhead
right.
F
C
It
comes
to
chatty,
and
so
so
know
that
on
api
has
an
additional
overhead,
even
if
you
use
templated
functions
because
you
still
have
to
you
still
have
to
like
create
some
objects.
C
E
C
Yeah
yeah
they're,
using
it
with,
I
think
it's
called
babylon
native.
If
you
remember
back
back
in
the
day
when
we
separated
out
the
js
native
api,
that
was
that
was
sort
of
a
nudge
given
to
us
from
a
person
at
microsoft
who
who
were
working
on
on
v8
only
instead
of
node.js
right
and
so
so
v8
only.
C
You
cannot
implement
node.js
apis,
but
you
still
can
use
an
api,
and
so
so
that
seems
to
have
matured
into
into
this
into
this
sort
of
back-end
stack
for
running
compute,
stuff
on
gpus
and
stuff
like
that,
but
but
yeah
they're
using
an
api.
So
so
that's
pretty
good.
B
Yeah
and
if
you
go
on
babylon
native
under
the
balances
they
now
implemented
the
nabi
for
javascript,
core
and
yeah,
that
is
the
engine
yeah
for
for
from
maple
no
yeah
ios,
yes
for
ios
and
yeah,
I
don't
know
if
they
had
something
about
spider
monk.
I
don't
remember
it's.
B
C
Possible
spider-monkey
they're,
going
at
least
from
what
I
gather
from
from
the
other
meeting
they're
going
the
wasm
route,
so
so
they're
probably
going
to
go
wash
them
rather
than
an
api.
But
we'll
see.
B
Yeah
so
yeah
they
implemented
the
four
yeah
an
api
for
different
javascript
engine.
It's
very
good,
so.
C
A
B
It's
just
a
core.
I
can
find
only
just
the
core
binding
here,
because
they.
F
B
Reactions
yes
as
v8,
because
on
android
they
use
v8,
so
essentially
yeah
yeah
yeah,
the
the
only
only
only
the
the
binding
for
javascript
core
now.
C
B
Yeah,
I'm
I'm
experimenting
with
quick
js
that
it's
another
just
angie.
I
won't
try
to
to
see
I
I
I
I
can
do
these
by
my
own,
just
like
a
exercise
and
see
where
I
can
go.
C
Yeah
yeah
yeah
yeah,
it's
a
lot
of
fun.
I
remember
when
I
implemented
it
for
for
for
jerryscript,
that's
not
the
implementation
that
ended
up
being
used,
so
the
current
implementation
is
in
javascript
is
by
somebody
else,
but
I
remember
having
lots
of
fun
tracking
the
data
and
then
setting
up
the
environment
and
all
that.
So
it's
a
lot
of
fun.